The 20kg 50mH chokes I have in cellar with 73 mOhm DC resistance are to large for the chassis.
I tried again with ear and I think 3 cm was the right distance to be able to hear the noise. I may damage dome if I try to get closer. But it seems I hear better than my DMM.
I did a hear test recently and the doctor looked at the curves and claimed I had a hearing like a 10 years old child.
Maybe MJ can redesign the filter for these kind of amps for higher voltage and current.
About temperature at heatsink. 56 C after a couple of hours.
With only one channel the subjective observation is that it plays nice and clean.
I tried again with ear and I think 3 cm was the right distance to be able to hear the noise. I may damage dome if I try to get closer. But it seems I hear better than my DMM.
I did a hear test recently and the doctor looked at the curves and claimed I had a hearing like a 10 years old child.
Maybe MJ can redesign the filter for these kind of amps for higher voltage and current.
About temperature at heatsink. 56 C after a couple of hours.
With only one channel the subjective observation is that it plays nice and clean.
There is very little hiss sound but hidden in this noise there is a little bit of ripple noise like a 100 Hz "dirty" noise. Not a clean 100 Hz.
This is probably from the SMPS ripple. I tried to put a 10uF cap across the PSU terminals but that changed noting. I think the only way to reduce this ripple is a serial choke. But the noise is less than my DMM can measure. Speakers are 94 dB and if you need to concentrate to hear it with ear almost inside tweeter then noise is very low. VFET amp has only "hiss". Will see if other amp behaves the same.
I have no Earth in my mains outlets but I build everything like there was. Chassis will be connected together via the DC-blocker or regenerator. Maybe if I had mains with earth it could reduce noise further. I only have outlets with earth in the kitchen.
This is probably from the SMPS ripple. I tried to put a 10uF cap across the PSU terminals but that changed noting. I think the only way to reduce this ripple is a serial choke. But the noise is less than my DMM can measure. Speakers are 94 dB and if you need to concentrate to hear it with ear almost inside tweeter then noise is very low. VFET amp has only "hiss". Will see if other amp behaves the same.
I have no Earth in my mains outlets but I build everything like there was. Chassis will be connected together via the DC-blocker or regenerator. Maybe if I had mains with earth it could reduce noise further. I only have outlets with earth in the kitchen.
The woofer and midrange speaker are silent so what I hear in tweeter could also be lower harmonics of the 100 kHz ripple noise or upper harmonics of 100 Hz. A clean 100 Hz should not be detectable in tweeter but I remember from also playing with conventional PSU's that the 100 Hz ripple noise was only audible in the tweeter. I think it is because it is a very "dirty" 100 Hz. Levels this low......maybe it is the ear which is much more sensibel for the higher harmonics.
move your ears together with your butttt in listening position and it'll be good for now 
post some niice pic of your build inside, so we can see is there anything else inside to route differently
post some niice pic of your build inside, so we can see is there anything else inside to route differently
Amps sounds very good with good "punch" so no problem with sound 🙂
There is just a bit more noise than e.g. VFET amp which uses same kind of SMPS but it sits in its own chassis. It drives both channels so current draw about same (3.2A). From memory ripple noise was lower. About 10mV or so. Why I measure so much more in Lazy SMPS (130 mV) I don't know. There is a little bit of hum from all speaker units if ear is close. Not audible in listening position.
Looking at pictures my own comment would be that boards could be arranged different to have much shorter signal wire to RCA connector. It is the most obvious weakness as I see it. But if I cut the wire and grounded at PCB it would not cure the noise.
What I concentrated on in the build was to have short distance between boards and also that the SMPS was out of the way from boards. So it ended as is.
The SMPS filters are mostly to reduce high freq? .....so will probably not help in my situation?
What I would like to try was to remove signal Gnd from chassis/earth just the test shortly if it made a difference?
If it did then I would now where the problem was?
There is just a bit more noise than e.g. VFET amp which uses same kind of SMPS but it sits in its own chassis. It drives both channels so current draw about same (3.2A). From memory ripple noise was lower. About 10mV or so. Why I measure so much more in Lazy SMPS (130 mV) I don't know. There is a little bit of hum from all speaker units if ear is close. Not audible in listening position.
Looking at pictures my own comment would be that boards could be arranged different to have much shorter signal wire to RCA connector. It is the most obvious weakness as I see it. But if I cut the wire and grounded at PCB it would not cure the noise.
What I concentrated on in the build was to have short distance between boards and also that the SMPS was out of the way from boards. So it ended as is.
The SMPS filters are mostly to reduce high freq? .....so will probably not help in my situation?
What I would like to try was to remove signal Gnd from chassis/earth just the test shortly if it made a difference?
If it did then I would now where the problem was?
Attachments
A comment more:
The SMPS is mounted via its heatsink at the bottom via six M4 screws. The heatsink is internal connected to earth. As you can see I also have a chassis screw just beside the SMPS. In principle are there two earth connections. One via the screw and one via the SMPS heatsink just beside each other.
The SMPS is mounted via its heatsink at the bottom via six M4 screws. The heatsink is internal connected to earth. As you can see I also have a chassis screw just beside the SMPS. In principle are there two earth connections. One via the screw and one via the SMPS heatsink just beside each other.
fiddle with takeout point of audio GND to chassis
one way - ditto from switcher positive rail screw
second way - ditto from SIT Drain screw
try both with NTC and direct (NTC shorted)
in any case, audio ground must be connected to chassis
one way - ditto from switcher positive rail screw
second way - ditto from SIT Drain screw
try both with NTC and direct (NTC shorted)
in any case, audio ground must be connected to chassis
For at short experiment it is ok to remove GND to chassis?
Safety wise I have no Earth in mains so......but there is a HPFI relay which will protect.
What I would try first is to remove NTC connection from positive rail (very short experiment). It is an easy test when amp is cooled down a bit.
SIT Drain need in all cases connection to PSU V+ (GND). Do you mean a wire direct from SIT Drain to NTC/Earth (instead of from PSU)?
I looked into PSU box for VFET amp. In this I have not Earth screw. The four screws in the corners of SMPS have chassis contact and contact to yellow/green wire from mains. Then I have NTC to one of these corner screws which goes to screen of DC power wire which connects to VFET amp (the 4 pole connector).
I could try something similar......to remove the chassis screw I use for mains earth and then use a corner of SMPS board. Then only heatsink of SMPS will have chassis contact?
The two 4 pole caps in serial in VFET PSU reduced ripple with a couple of mV......from 11 to 9 or so. They just act as sophisticated wires.
There must be a reason why I see to much more ripple.
Safety wise I have no Earth in mains so......but there is a HPFI relay which will protect.
What I would try first is to remove NTC connection from positive rail (very short experiment). It is an easy test when amp is cooled down a bit.
SIT Drain need in all cases connection to PSU V+ (GND). Do you mean a wire direct from SIT Drain to NTC/Earth (instead of from PSU)?
I looked into PSU box for VFET amp. In this I have not Earth screw. The four screws in the corners of SMPS have chassis contact and contact to yellow/green wire from mains. Then I have NTC to one of these corner screws which goes to screen of DC power wire which connects to VFET amp (the 4 pole connector).
I could try something similar......to remove the chassis screw I use for mains earth and then use a corner of SMPS board. Then only heatsink of SMPS will have chassis contact?
The two 4 pole caps in serial in VFET PSU reduced ripple with a couple of mV......from 11 to 9 or so. They just act as sophisticated wires.
There must be a reason why I see to much more ripple.
Attachments
The earth connection if it is there or not has nothing to do with PSU ripple. It stays around 130 mV during all my experiments with amp on bench with RCA shorted.
I assume that is just how it is. I can ask Connex if that is expected. I can also power VFET amp up and measure again so I can compare the two PSU's to be sure I remember correct. I think my lab supply has at least as much ripple so it would probably make no sense to try that insted for one of the amps. It could be fun just to make something that could reduce ripple to be sure that is the problem. Now the problem is not huge but could be nice to have it as good as my other amps regarding noise.
I assume that is just how it is. I can ask Connex if that is expected. I can also power VFET amp up and measure again so I can compare the two PSU's to be sure I remember correct. I think my lab supply has at least as much ripple so it would probably make no sense to try that insted for one of the amps. It could be fun just to make something that could reduce ripple to be sure that is the problem. Now the problem is not huge but could be nice to have it as good as my other amps regarding noise.
re-read post #309
that's what you can do, regarding relation of audio GND to case
now, few facts:
-switcher case is connected to chassis
-both rails are floating and need to be, up to the moment where you connect ( in some way) audio GND to chassis (yes, two ways - either direct or though some NTC whatever gizmo)
-negative rail stays free to roam and do its job
-point from where you're taking audio GND to send it to chassis is - either ditto from PSU screw of positive rail ( that's bloody audio GND)......... or from SIT Drain screw
I think I exhausted all available ways of writing same thing
that's what you can do, regarding relation of audio GND to case
now, few facts:
-switcher case is connected to chassis
-both rails are floating and need to be, up to the moment where you connect ( in some way) audio GND to chassis (yes, two ways - either direct or though some NTC whatever gizmo)
-negative rail stays free to roam and do its job
-point from where you're taking audio GND to send it to chassis is - either ditto from PSU screw of positive rail ( that's bloody audio GND)......... or from SIT Drain screw
I think I exhausted all available ways of writing same thing
Yes, but I am quite almost sure I need lower ripple to get lower noise at output.
I will reconnect the chassis earth connection. I will do it so PSU only have chassis connection via heatsink and not via chassis screw and heatsink. Then I am sure there will be no ground loop there at least.
I tried to disconnect amp from PSU and used a test speaker load (2 x 8 ohm resistors) to draw 4A. Then I see 148 mV ripple.
Maybe it was an error to get a smaller transformer for my purpose. VFET PSU has the standard one so it is a 2kW module. Then the logic says that if you only use 120W then ripple is small compared to use full power. But Connex people tried to convince me about that is was an advantage regarding noise to reduce max. power using a smaller transformer.
Generel speaking.....is 130 mV ripple a lot for an audio SE power amp PSU?
I made a PSU for a tube amp where the design goal was 1 mV ripple and the same for M2X that has a CRCLC configuration.
Then when I got SMPS for VFET I saw the 10 mV as much......but amp was silent and then I was happy.
I will reconnect the chassis earth connection. I will do it so PSU only have chassis connection via heatsink and not via chassis screw and heatsink. Then I am sure there will be no ground loop there at least.
I tried to disconnect amp from PSU and used a test speaker load (2 x 8 ohm resistors) to draw 4A. Then I see 148 mV ripple.
Maybe it was an error to get a smaller transformer for my purpose. VFET PSU has the standard one so it is a 2kW module. Then the logic says that if you only use 120W then ripple is small compared to use full power. But Connex people tried to convince me about that is was an advantage regarding noise to reduce max. power using a smaller transformer.
Generel speaking.....is 130 mV ripple a lot for an audio SE power amp PSU?
I made a PSU for a tube amp where the design goal was 1 mV ripple and the same for M2X that has a CRCLC configuration.
Then when I got SMPS for VFET I saw the 10 mV as much......but amp was silent and then I was happy.
only your satisfaction is measure of that
you know that simple SE constructions are having lousy ripple rejection figures, and you had best intentions when ordering switchers
now, only if you can avoid pushing your ear in cones, you could be happy
one day, when I grow up, I'm going to finish my own SB monoblocks, for which I already have ready 9mH chokes
MEPER, As ZM mentioned, the biggest contributor to the noise is the high PS ripple, and the easy solution is to add an LC section to your power supply. You don't need a 20kg 50mH choke. Hammond makes the 159ZJ which Mouser sells for 35.68 Euro each, and according to their website shipping is free for orders over 50 Euro, if you pay in Euro. The 159ZJ is a bit over 1 kg in weight: Hammond 159ZJ
I use CLC comprising 44mF - 159ZJ - 44mH in my single ended SIT amplifiers, and I hear absolutely no noise from my 103dB speakers.
I use CLC comprising 44mF - 159ZJ - 44mH in my single ended SIT amplifiers, and I hear absolutely no noise from my 103dB speakers.
I may have space for that choke inside the chassis but 44mF 100V is very large / tall. Hopefully a smaller one can do it in my case. The one I use for output cap is 22 mF / 100V......one more of these could probably be in there.
I plan for using Drain also for the chassis via NTC and I select a chassis point very close to the SIT. I guess that is ok that the heatsink panel the SIT is sitting on is also the chassis point?
A re-measurement showed that PSU heatsink is not connected to earth. I have fooled myself regarding this earlier. Then Drain of SIT becomes a real star-point for the amp. I will probably go for the 159ZJ for a future upgrade and a decent cap. Maybe it was the same I used for M2X. It was also a Hammond 159 type but maybe less mH, less Ohm and more A.
I plan for using Drain also for the chassis via NTC and I select a chassis point very close to the SIT. I guess that is ok that the heatsink panel the SIT is sitting on is also the chassis point?
A re-measurement showed that PSU heatsink is not connected to earth. I have fooled myself regarding this earlier. Then Drain of SIT becomes a real star-point for the amp. I will probably go for the 159ZJ for a future upgrade and a decent cap. Maybe it was the same I used for M2X. It was also a Hammond 159 type but maybe less mH, less Ohm and more A.
You probably could get by with much less than 44mF. My CLC of 44mF - 159ZJ - 44mF saw nearly 65V of ripple.
It would be interesting to see a FFT of your amplifier's output (with a 1W signal for reference), to see the noise.
It would be interesting to see a FFT of your amplifier's output (with a 1W signal for reference), to see the noise.
my praxis is - chassis screw close to IEC, all wires going to chassis on that one screw
not saying that's only way but works for me, and one habitual thing keeping me from thinking too much
22mF as final filter cap is certainly enough after choke
not saying that's only way but works for me, and one habitual thing keeping me from thinking too much
22mF as final filter cap is certainly enough after choke
Ok, I will order chokes and 22mF caps and rearrange chassis screw close to IEC and then continue when chokes/caps arrives.
Will see if I get time for measurement on one amp. Vacation over now 🙂
Also need to clean-up.......before works starts.
What I did notice when looking a PSU AC output with scope was more noise than expected. Both HF and LF noise. The LC should clean some of this.
When works starts I don't have much time to "play".
Will see if I get time for measurement on one amp. Vacation over now 🙂
Also need to clean-up.......before works starts.
What I did notice when looking a PSU AC output with scope was more noise than expected. Both HF and LF noise. The LC should clean some of this.
When works starts I don't have much time to "play".